What's New with SAS Model Manager? pzmm (Python Zip Model Management) module updates

[#1]

Do you create models in Python and want to leverage SAS Model Manager to govern, deploy, and/or monitor your models? If yes, you'll want to check out pzmm and the videos below.

pzmm (also known as Python Zip Model Management) is a sasctl module created and maintained on GitHub by SAS Model Manager R&D. The package enables users of SAS Model Manager on SAS Viya and SAS Open Model Manager to zip through the process of importing Python models into the common model repository. In order to facilitate model imports, the module allows completion of the following tasks:

• Writes JSON files to read in the model information including:
  • fileMetadata.json - specifies the file roles for the names of the input and output variables files, the Python score code file, and the Python pickle file
  • ModelProperties.json - used to set the model properties that are read during the import process
  • inputVar.json and outputVar.json - used to set the input and output variables of the model
  • dmcas_fitstat.json - optional file that provides the fit statistics that are associated with the imported model, which are either user-generated or data-generated
  • dmcas_lift.json and dmcas_roc.json - optional files that provide the Lift and ROC plots associated with the imported model, which are data-generated
• Writes the *score.py model file used for model scoring
• Serializes a trained model into a binary pickle file
• Archives all relevant model files into a ZIP file and imports the model using REST API calls

Please check out this video where ScottLindauer walks through the use of pzmm with SAS Model Manager.

pzmm Explained

Here is a link to the notebook ScottLindauer used in this demo.

For more information about how SAS Model Manager supports models developed in Python and other languages, please check out the article: Organize and manage all types of analytic models and pipelines. For more information about registering Python models like scikit-learn, TensorFlow, and XGBoost using sasctl and PZMM module see these links below:

• sasctl - https://github.com/sassoftware/python-sasctl 
• pzmm - https://github.com/sassoftware/python-sasctl/tree/master/src/sasctl/pzmm

Now that you've successfully registered your Python model in SAS Model Manager, see how you can easily compare and contrast your model with all types of models directly within SAS Model Manager.

Quick SAS Model Manager Demo

Now that your Python model is deployed, see an example of using other consumable SAS Viya APIs. This demo shows how users can easily leverage automated modeling and automated decisioning, all managed by SAS Model Manager, using a custom application and SAS Viya APIs.

Consumable SAS Viya APIs Demo

To learn how to create and build your machine learning web application using SAS AutoML, check out this great SAS Users blog post by @paugre. For additional resources, please check out:

• SAS Visual Data Mining & Machine Learning
• Automated Machine Learning Pipelines
• Easier Feature Generation and New Data Science Pilot Actions
• An Easy Button for Data Science in Python 

Interested to learn about other SAS Model Manager features? Please check out other "What's new with SAS Model Manager?" posts including:

• New/updated Model Management Macros generating JSON files
• Simple filtering dashboards plus enhanced search options
• Customize SAS Model Manager performance monitoring reports with embedded web components